LAMARCK1ANA SEMlGIGAS. 553 



We have since found small and large chromosomes in the nucleus 

 of 0. Lamarckiana itself. 



It is well known that the heterogamic mutations of 0. Lamarckiana 

 as a rule are connected with the doubling of one of the chromo- 

 somes. This brings their number in the vegetative cells up to 15. 

 Why this should be so we do not know, since the doubling may be 

 accompanied, for each chromosome, with widely different types of 

 mutations, such as albida and lata in one case, or oblonga and scin- 

 tillans in another. Moreover the hereditary features are not deter- 

 mined by this doubling, since in almost every group sesquiplex or 

 constant forms are found besides the typical dimorphic or splitting 

 ones. The connection is such, however, that in ordinary cases one 

 may confidently rely upon it for conclusions. 



It would follow that the progeny of mutants of the type of semi- 

 gigas might afford a good test for our conceptions. It is astonishingly 

 rich in forms, some of which repeat the old types (0. lata, 0. scin- 

 tillans and 0. oblonga) in a relatively high percentage, whereas 

 others constitute new types occurring in smaller numbers or even 

 in single individuals, as has been described by one of us (4). At that 

 time (1913), however, only three of the dimorphic prototypes (0. 

 lata, 0. scintillans, and 0. spathulata) had been discovered, and thus 

 the principle for grouping the numerous forms failed. 0. cana, 0. 

 pallescens, and 0. liquida have since been distinguished (5), com- 

 pleting the series. If we apply these forms as models for distinction 

 to the motley progeny of semigigas, the main groups at once become 

 prominent. Besides lata and scintillans, which are always easily 

 recognized, cana and spathulata become evident, and palles- 

 cens and liquida can also be distinguished, although not with 

 the same clearness. In some specimens these types seemed to 

 be as pure as those of the corresponding old races, but in many 

 others they were evidently obscured by subordinate deviating 

 characters. 



From numerous cultures of the progeny of different types of 

 semigigas, we discovered that the clearness of the groups depends 

 largely on the choice of the type which yields the pollen for the 

 fertilization. In order to get a sufficient supply of good seeds the 

 flowers of semigigas have to be fecundated by means of other pollen; 

 self-fertilization produces too few germinating seeds. Some pollen 

 proved to be dominant over a majority of the mutations, making 

 the distinction very difficult and in some degree uncertain (as that 

 of 0. Lama r ckiana mut. gigas); other types proved to be recessive, 



